7 |
use lib './lib'; |
use lib './lib'; |
8 |
|
|
9 |
use WebPAC::Common 0.02; |
use WebPAC::Common 0.02; |
10 |
use WebPAC::Parser 0.07; |
use WebPAC::Parser 0.08; |
11 |
use WebPAC::Input 0.13; |
use WebPAC::Input 0.16; |
12 |
use WebPAC::Store 0.11; |
use WebPAC::Store 0.14; |
13 |
use WebPAC::Normalize 0.22; |
use WebPAC::Normalize 0.22; |
14 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
15 |
use WebPAC::Validate 0.06; |
use WebPAC::Validate 0.11; |
16 |
use WebPAC::Output::MARC; |
use WebPAC::Output::MARC; |
17 |
use WebPAC::Config; |
use WebPAC::Config; |
18 |
use Getopt::Long; |
use Getopt::Long; |
21 |
use File::Slurp; |
use File::Slurp; |
22 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
23 |
use Storable qw/dclone/; |
use Storable qw/dclone/; |
24 |
|
use Pod::Usage qw/pod2usage/; |
25 |
|
|
26 |
use Proc::Queue size => 1; |
use Proc::Queue size => 1; |
27 |
use POSIX ":sys_wait_h"; # imports WNOHANG |
use POSIX ":sys_wait_h"; # imports WNOHANG |
32 |
|
|
33 |
B<this command will probably go away. Don't get used to it!> |
B<this command will probably go away. Don't get used to it!> |
34 |
|
|
35 |
Options: |
=head1 OPTIONS |
36 |
|
|
37 |
=over 4 |
=over 4 |
38 |
|
|
68 |
|
|
69 |
turn on extra validation of imput records, see L<WebPAC::Validation> |
turn on extra validation of imput records, see L<WebPAC::Validation> |
70 |
|
|
71 |
|
You can use special variables C<$database> and $C<$input> in this parametar |
72 |
|
like C<--validate 'conf/validate/$database-$input'> to construct filename |
73 |
|
|
74 |
|
=item --validate-delimiters path/to/validate_delimiters_file |
75 |
|
|
76 |
|
this option is used with C<--validate> to turn on extra validation of |
77 |
|
delimiters. If file is non existant, it will be created on first run. |
78 |
|
|
79 |
|
=item --marc-generate |
80 |
|
|
81 |
|
Generate MARC file. This will automatically be on if file contains C<marc*> directives. |
82 |
|
You can use this option as C<--no-marc-generate> to disable MARC generation. |
83 |
|
|
84 |
=item --marc-lint |
=item --marc-lint |
85 |
|
|
86 |
By default turned on if normalisation file has C<marc*> directives. You can disable lint |
By default turned on if normalisation file has C<marc*> directives. You can disable lint |
116 |
my $only_filter; |
my $only_filter; |
117 |
my $stats = 0; |
my $stats = 0; |
118 |
my $validate_path; |
my $validate_path; |
119 |
|
my $validate_delimiters_path; |
120 |
|
my $marc_generate = 1; |
121 |
my $marc_lint = 1; |
my $marc_lint = 1; |
122 |
my $marc_dump = 0; |
my $marc_dump = 0; |
123 |
my $parallel = 0; |
my $parallel = 0; |
124 |
my $only_links = 0; |
my $only_links = 0; |
125 |
my $merge = 0; |
my $merge = 0; |
126 |
|
my $help; |
127 |
|
|
128 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $log = _new WebPAC::Common()->_get_logger(); |
129 |
|
|
137 |
"debug+" => \$debug, |
"debug+" => \$debug, |
138 |
"stats" => \$stats, |
"stats" => \$stats, |
139 |
"validate=s" => \$validate_path, |
"validate=s" => \$validate_path, |
140 |
|
"validate-delimiters=s" => \$validate_delimiters_path, |
141 |
|
"marc-generate!" => \$marc_generate, |
142 |
"marc-lint!" => \$marc_lint, |
"marc-lint!" => \$marc_lint, |
143 |
"marc-dump!" => \$marc_dump, |
"marc-dump!" => \$marc_dump, |
144 |
"parallel=i" => \$parallel, |
"parallel=i" => \$parallel, |
145 |
"only-links!" => \$only_links, |
"only-links!" => \$only_links, |
146 |
"merge" => \$merge, |
"merge" => \$merge, |
147 |
|
"help" => \$help, |
148 |
); |
); |
149 |
|
|
150 |
|
$marc_generate = 0 if ( $validate_delimiters_path ); |
151 |
|
|
152 |
|
pod2usage(-verbose => 2) if ($help); |
153 |
|
|
154 |
my $config = new WebPAC::Config( path => $config_path ); |
my $config = new WebPAC::Config( path => $config_path ); |
155 |
|
|
156 |
#print "config = ",dump($config) if ($debug); |
#print "config = ",dump($config) if ($debug); |
159 |
|
|
160 |
$log->info( "-" x 79 ); |
$log->info( "-" x 79 ); |
161 |
|
|
162 |
|
my $log_file = 'log'; |
163 |
|
|
164 |
|
if (-e $log_file ) { # && -s $log_file > 5 * 1024 * 1024) { |
165 |
|
$log->info("moved old log with ", -s $log_file, " bytes to '${log_file}.old'"); |
166 |
|
rename $log_file, "${log_file}.old" || $log->logwarn("can't rename $log_file to ${log_file}.old: $!"); |
167 |
|
} |
168 |
|
|
169 |
my $estcmd_fh; |
my $estcmd_fh; |
170 |
my $estcmd_path = './estcmd-merge.sh'; |
my $estcmd_path = './estcmd-merge.sh'; |
175 |
$log->info("created merge batch file $estcmd_path"); |
$log->info("created merge batch file $estcmd_path"); |
176 |
} |
} |
177 |
|
|
|
|
|
178 |
my $validate; |
my $validate; |
179 |
$validate = new WebPAC::Validate( |
$validate = new WebPAC::Validate( |
180 |
path => $validate_path, |
delimiters => $config->webpac('delimiters'), |
181 |
) if ($validate_path); |
) if ($validate_path || $validate_delimiters_path); |
|
|
|
182 |
|
|
183 |
my $use_indexer = $config->use_indexer; |
my $use_indexer = $config->use_indexer; |
184 |
|
$stats ||= $validate; |
185 |
if ($stats) { |
if ($stats) { |
186 |
$log->debug("option --stats disables update of indexing engine..."); |
$log->debug("disabled indexing for stats collection"); |
187 |
$use_indexer = undef; |
$use_indexer = undef; |
188 |
} else { |
} else { |
189 |
$log->info("using $use_indexer indexing engine..."); |
$log->info("using $use_indexer indexing engine..."); |
212 |
return $c; |
return $c; |
213 |
} |
} |
214 |
|
|
215 |
while (my ($database, $db_config) = each %{ $config->databases }) { |
foreach my $database ( sort keys %{ $config->databases } ) { |
216 |
|
my $db_config = $config->databases->{$database}; |
217 |
|
|
218 |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
219 |
next if ($only_database && $database !~ m/$only_database/i); |
next if ($only_database && $database !~ m/$only_database/i); |
228 |
} |
} |
229 |
|
|
230 |
my $indexer; |
my $indexer; |
231 |
if ($use_indexer) { |
if ($use_indexer && $parser->have_rules( 'search', $database )) { |
232 |
|
|
233 |
my $cfg_name = $use_indexer; |
my $cfg_name = $use_indexer; |
234 |
$cfg_name =~ s/\-.*$//; |
$cfg_name =~ s/\-.*$//; |
380 |
my $rec = shift || die "need rec!"; |
my $rec = shift || die "need rec!"; |
381 |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
my $mfn = $rec->{'000'}->[0] || die "need mfn in 000"; |
382 |
|
|
|
$store->save_row( |
|
|
database => $database, |
|
|
input => $input_name, |
|
|
id => $mfn, |
|
|
row => $rec, |
|
|
); |
|
|
|
|
383 |
WebPAC::Normalize::data_structure( |
WebPAC::Normalize::data_structure( |
384 |
row => $rec, |
row => $rec, |
385 |
rules => $rules, |
rules => $rules, |
403 |
lookup_coderef => $lookup_coderef, |
lookup_coderef => $lookup_coderef, |
404 |
lookup => $lookup_jar, |
lookup => $lookup_jar, |
405 |
%{ $input }, |
%{ $input }, |
406 |
|
load_row => sub { |
407 |
|
my $a = shift; |
408 |
|
return $store->load_row( |
409 |
|
database => $database, |
410 |
|
input => $input_name, |
411 |
|
id => $a->{id}, |
412 |
|
); |
413 |
|
}, |
414 |
|
save_row => sub { |
415 |
|
my $a = shift; |
416 |
|
return $store->save_row( |
417 |
|
database => $database, |
418 |
|
input => $input_name, |
419 |
|
id => $a->{id}, |
420 |
|
row => $a->{row}, |
421 |
|
); |
422 |
|
}, |
423 |
|
|
424 |
); |
); |
425 |
|
|
426 |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
my $lookup_data = WebPAC::Normalize::_get_lookup(); |
427 |
|
|
428 |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
if (defined( $lookup_data->{$database}->{$input_name} )) { |
429 |
$log->debug("created following lookups: ", dump( $lookup_data )); |
$log->debug("created following lookups: ", sub { dump( $lookup_data ) } ); |
430 |
|
|
431 |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
foreach my $key (keys %{ $lookup_data->{$database}->{$input_name} }) { |
432 |
$store->save_lookup( |
$store->save_lookup( |
446 |
print $report_fh "Report for database '$database' input '$input_name' records ", |
print $report_fh "Report for database '$database' input '$input_name' records ", |
447 |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
$offset || 1, "-", $limit || $input->{limit} || $maxmfn, "\n\n"; |
448 |
$log->info("Generating report file $path"); |
$log->info("Generating report file $path"); |
449 |
|
|
450 |
|
if ( $validate ) { |
451 |
|
$validate->read_validate_file( $validate->fill_in( $validate_path, database => $database, input => $input_name ) ); |
452 |
|
$validate->read_validate_delimiters_file( $validate->fill_in( $validate_delimiters_path, database => $database, input => $input_name ) ); |
453 |
|
} |
454 |
} |
} |
455 |
|
|
456 |
my $marc; |
my $marc; |
457 |
if ($parser->generate_marc( $database, $input_name )) { |
if ($marc_generate && $parser->have_rules( 'marc', $database, $input_name )) { |
458 |
$marc = new WebPAC::Output::MARC( |
$marc = new WebPAC::Output::MARC( |
459 |
path => "out/marc/${database}-${input_name}.marc", |
path => "out/marc/${database}-${input_name}.marc", |
460 |
lint => $marc_lint, |
lint => $marc_lint, |
482 |
foreach my $db (keys %$depends) { |
foreach my $db (keys %$depends) { |
483 |
foreach my $i (keys %{$depends->{$db}}) { |
foreach my $i (keys %{$depends->{$db}}) { |
484 |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
foreach my $k (keys %{$depends->{$db}->{$i}}) { |
485 |
|
my $t = time(); |
486 |
$log->debug("loading lookup $db/$i"); |
$log->debug("loading lookup $db/$i"); |
487 |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
$lookup_hash->{$db}->{$i}->{$k} = $store->load_lookup( |
488 |
database => $db, |
database => $db, |
489 |
input => $i, |
input => $i, |
490 |
key => $k, |
key => $k, |
491 |
); |
); |
492 |
|
$log->debug(sprintf("lookup $db/$i took %.2fs", time() - $t)); |
493 |
} |
} |
494 |
} |
} |
495 |
} |
} |
496 |
|
|
497 |
$log->debug("lookup_hash = ", dump( $lookup_hash )); |
$log->debug("lookup_hash = ", sub { dump( $lookup_hash ) }); |
498 |
} |
} |
499 |
|
|
500 |
|
|
502 |
|
|
503 |
my $row = $input_db->fetch || next; |
my $row = $input_db->fetch || next; |
504 |
|
|
505 |
|
$total_rows++; |
506 |
|
|
507 |
my $mfn = $row->{'000'}->[0]; |
my $mfn = $row->{'000'}->[0]; |
508 |
|
|
509 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
if (! $mfn || $mfn !~ m#^\d+$#) { |
514 |
|
|
515 |
|
|
516 |
if ($validate) { |
if ($validate) { |
517 |
if ( my $errors = $validate->validate_errors( $row, $input_db->dump ) ) { |
if ( my $errors = $validate->validate_rec( $row, $input_db->dump_ascii ) ) { |
518 |
$log->error( "MFN $mfn validation error:\n", |
$log->error( "MFN $mfn validation error:\n", |
519 |
$validate->report_error( $errors ) |
$validate->report_error( $errors ) |
520 |
); |
); |
521 |
} |
} |
522 |
|
next; # validation doesn't create any output |
523 |
} |
} |
524 |
|
|
525 |
my $ds = WebPAC::Normalize::data_structure( |
my $ds = WebPAC::Normalize::data_structure( |
538 |
}, |
}, |
539 |
); |
); |
540 |
|
|
541 |
$log->debug("ds = ",dump($ds)); |
$log->debug("ds = ", sub { dump($ds) }) if ($ds); |
542 |
|
|
543 |
$store->save_ds( |
$store->save_ds( |
544 |
database => $database, |
database => $database, |
560 |
$marc->add( |
$marc->add( |
561 |
id => $mfn . ( $i ? "/$i" : '' ), |
id => $mfn . ( $i ? "/$i" : '' ), |
562 |
fields => $fields, |
fields => $fields, |
563 |
leader => WebPAC::Normalize::marc_leader(), |
leader => WebPAC::Normalize::_get_marc_leader(), |
564 |
row => $row, |
row => $row, |
565 |
); |
); |
566 |
$i++; |
$i++; |
568 |
|
|
569 |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
$log->info("Created $i instances of MFN $mfn\n") if ($i > 1); |
570 |
} |
} |
|
|
|
|
$total_rows++; |
|
571 |
} |
} |
572 |
|
|
573 |
if ($validate) { |
if ($validate) { |
576 |
$log->info("validation errors:\n$errors\n" ); |
$log->info("validation errors:\n$errors\n" ); |
577 |
print $report_fh "$errors\n" if ($report_fh); |
print $report_fh "$errors\n" if ($report_fh); |
578 |
} |
} |
579 |
|
|
580 |
|
print $report_fh "\nAll possible subfields/delimiter templates:\n", $validate->delimiters_templates( report => 1, current_input => 1 ), "\n\n"; |
581 |
|
|
582 |
|
# must be last thing that touches $validate for this input |
583 |
|
$validate->reset; |
584 |
} |
} |
585 |
|
|
586 |
if ($stats) { |
if ($stats) { |
593 |
$marc->finish if ($marc); |
$marc->finish if ($marc); |
594 |
|
|
595 |
# close report |
# close report |
596 |
close($report_fh) if ($report_fh) |
close($report_fh) if ($report_fh); |
|
|
|
597 |
} |
} |
598 |
|
|
599 |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
eval { $indexer->finish } if ($indexer && $indexer->can('finish')); |
620 |
$log->info("all parallel processes finished"); |
$log->info("all parallel processes finished"); |
621 |
} |
} |
622 |
|
|
623 |
|
# save new delimiters if needed |
624 |
|
$validate->save_delimiters_templates if ( $validate ); |
625 |
|
|
626 |
# |
# |
627 |
# handle links or merge after indexing |
# handle links or merge after indexing |
628 |
# |
# |